smtek/ Swift-Qwen3.8-27B:Q4_K_M

35 22 hours ago

Reasoning model focused on shorter, token-efficient thinking + DFlash2 + map-k4v

... /
params
b15f80218408 · 181B
{
"min_p": 0,
"num_batch": 512,
"num_ctx": 143360,
"num_predict": -1,
"presence_penalty": 0,
"repeat_penalty": 1,
"stop": [
"<|im_start|>",
"<|im_end|>"
],
"temperature": 0.6,
"top_k": 20,
"top_p": 0.95
}